The overall goal of this sequential mixed methods research is to adapt and test a culturally appropriate and effective peer group intervention for HIV prevention specifically for couples in Malawi. Married couples are at risk for HIV infection due to unprotected sex either before or after marriage, and a growing number of couples face the issue of discordant HIV status. However, no current HIV prevention interventions specifically address the unique HIV prevention needs of couples. Gender inequality has long been recognized as a barrier to HIV prevention for married couples, especially the wives. However, the Pi's recent dissertation research has found that the actual decision-making of couples in rural Malawi is more complex than a simple script where the husband always dominates. Couples used husband-dominant cultural scripts, but they also used three other cultural scripts to support shared decision-making: communicating openly, maintaining harmony, and children's welfare. Most couples used a mix of husband-dominated, wife dominated, and shared decision-making. In Phase I. these well-accepted cultural scripts and the shared decision-making style already widely used by couples will be integrated into an existing peer group intervention already used successfully in Malawi. This Couples Peer Group Intervention will be tailored for the needs of married couples. Measures of couple- specific outcomes will also be developed. In Phase II. the intervention will be piloted. A process evaluation will document fidelity, acceptability and feasibility of the intervention. An outcomes evaluation of short-term impacts of the intervention for married couples receiving the intervention compared to a delayed control group, using a pretest and a posttest at three months. The study will be conducted in four low income rural communities in two districts, one with a matrilineal and one with a patrilineal marriage system. Hypotheses will be tested using binomial tests of proportions for binary outcomes and analysis of variance with contrasts for continuous variables. Any baseline differences will be controlled for. We will recruit a total of 120 couples per group at baseline, and with an 80% retention rate the final sample will be at least 96 couples per group. With alpha=.05 this sample will give adequate power (.80) to identify an effect size of .35 for continuous variables (a medium effect). The proposed research is an innovative scientific advance that will integrate culturally appropriate couple decision-making cultural scripts and shared decision-making styles with peer group intervention based on behavioral change theory. The community-based participatory research and examination of both individual-level and couple-level outcomes is also innovative. An acceptable and effective HIV prevention for married couples will have high significance for global HIV prevention.